New contest

New contest

So the new challenge is started! This time it'll be slightly longer - 4 months.

Details: http://intel-software-academic-program.com/pages/contests

Good luck everyone, and have fun!

6 posts / 0 new
Last post
For more complete information about compiler optimizations, see our Optimization Notice.

Hey,

I see the contest has evolved ! This subject seems very interesting !

I didn't find the information but... can we participate to this contest with someone else to the contest like the previous ones ?

I'm joining eviltosha : good luck and have fun !

Timothé Viot, Engineer student, Insa Rennes 1 France

Hi there!

It is always nice to see old candidates participating one more time. To answer the question, on this edition, you can only participate alone. Even if your whole class join the contest with a teacher, each candidate must be alone.

Good luck to all the contestants and I wish you a lot of fun.

Hello, i am novice in this context, I would like to know, what soft can I use?
As I read from reference,you got server Xenon Phi, and got intel compiler there, OpenMP, TBB, MPI, Cilk.
I want to ask you few questions.
 - I have to compile with intel compiler only?
 - Is it posible to use Parallel studio for free as a Student? May be it is nesesery to use ParallelStudio? May be it is enpugth to use intel compiler? So, how can I use intel compiler for free more then one month?
 - What do you mean "We only accept native Xeon PHI code for this contest, no offloading" ? What is native Xeon PHI code?

Hi Alex and welcome to this contest !

You can use approximativly whatever you want, but you should use C/C++ because... well I'm not sure about the compatibility of other languages on the Xeon Phi but... well, the answer to your 3rd question will help you.

- You don't have to compile your code with Intel Compiler necessarly, gcc/g++ is enough but in fact, intel compiler offers usefull feature for vectorization (although remarks on loops are not very clear for me...). So the answer to your question is no. You can use gcc, g++ or any other appropriate compiler. (icc/icpc is just a way to have more informations on performances on your code, I think you could find some informations on the intel website).

- Yes, it is. I actually got a free version of the Parallel studio. See this page : http://software.intel.com/en-us/non-commercial-software-development. You will just have to complete a form and then, you will receive an email explaining how to install it. I had no problems on my debian computer.
If I remember well what I read and if you use this way to obtain the compiler, you will just not be able to update Intel Parallel Studio. I remember that it was one year, and not a month, but I may be wrong.

- If you want informations on Xeon Phi, how to use it and learning more about what you can do with it, you should visit this page : http://www.intel-software-academic-program.com/courses/. If you scroll down to HPC, you will find some interesting things on Xeon Phi. 

I hope I helped you ! :)
Regards 

Timothé Viot, Engineer student, Insa Rennes 1 France

In theory, you can compile with g++ but for the Xeon phi, it will not work.

Compiling for Xeon Phi is possible, but only with icpc right now. 

And actually, icpc do not just provide information on vectorization. It also works very well for auto vectorization in many case. 

In addition, you can use the tools available with Parallel Studio such as Vtune or Inspector. Vtune provides very interesting data to help you optimize the most consuming functions of your application. Inspector will help you to find race conditions and memory mistakes.

Login to leave a comment.